A ratio-based change detection method known as multiratio fusion (MRF) is proposed and tested. The MRF framework builds on other change detection components proposed in this work: dual ratio (DR) and multiratio (MR). The DR method involves two ratios coupled with adaptive thresholds to maximize detected changes and minimize false alarms. The use of two ratios is shown to outperform the single ratio case when the means of the image pairs are not equal. MR change detection builds on the DR method by including negative imagery to produce four total ratios with adaptive thresholds. Inclusion of negative imagery is shown to improve detection sensitivity and to boost detection performance in certain target and background cases. MRF further expands this concept by fusing together the ratio outputs using a routine in which detections must be verified by two or more ratios to be classified as a true changed pixel. The proposed method is tested with synthetically generated test imagery and real datasets with results compared to other methods found in the literature. DR is shown to significantly outperform the standard single ratio method. MRF produces excellent change detection results that exhibit up to a 22% performance improvement over other methods from the literature at low false-alarm rates.
Imagery from unmanned aerial systems (UAS) needs compression prior to transmission to a receiver for further processing. Once received, automated image exploitation algorithms, such as frame-to-frame registration, target tracking, and target identification, are performed to extract actionable information from the data. Unfortunately, in a compress-then-analyze system, exploitation algorithms must contend with artifacts introduced by lossy compression and transmission. Identifying metrics that enable compression engines to predict exploitation degradation could allow encoders the ability of tailoring compression for specific exploitation algorithms. This study investigates the impact of H.264 and JPEG2000 compression on target tracking through the use of a multi-hypothesis blob tracker. Used quality metrics include PSNR, VIF, and IW-SSIM.
We present a 3D change detection framework designed to support various applications in changing environmental conditions. Previous efforts have focused on image filtering techniques that manipulate the intensity values of the image to create a more controlled and unnatural illumination. Since most applications require detecting changes in a scene irrespective of the time of day and present lighting conditions, image filtering algorithms fail to suppress the illumination differences enough for Background Model (BM) subtraction to be effective. Our approach completely eliminates the illumination challenges from the change detection problem. The algorithm is based on our previous work in which we have shown a capability to reconstruct a surrounding environment in near real-time processing speeds. The algorithm, namely Dense Point-Cloud Representation (DPR), allows for a 3D reconstruction of a scene using only a single moving camera. In order to eliminate any effects of the illumination change, we convert each point-cloud model into a 3D binary voxel grid. A `1' is assigned to voxels containing points from the model while a `0' is assigned to voxels with no points. We detect the changes between the two environments by volumetrically subtracting the registered 3D binary voxel models. This process is extremely computationally efficient due to logic-based operations available when handling binary models. We evaluate the 3D change detection framework by experimenting on the same scene with aerial imagery captured at various times.
The use of hyperspectral imaging is a fast growing field with many applications in the civilian, commercial and military sectors. Hyperspectral images are typically composed of many spectral bands in the visible and infrared regions of the electromagnetic spectrum and have the potential to deliver a great deal of information about a remotely sensed scene. One area of interest regarding hyperspectral images is anomaly detection, or the ability to find spectral outliers within a complex background in a scene with no a priori information about the scene or its specific contents. Anomaly detectors typically operate by creating a statistical background model of a hyperspectral image and measuring anomalies as image pixels that do not conform properly to that given model. In this study we compare the performance over diurnal and seasonal changes for several different anomaly detection methods found in the literature and a new anomaly detector that we refer to as the fuzzy cluster-based anomaly detector. Here we also compare the performance of several anomaly-based change detection algorithms. Our results indicate that all anomaly detectors tested in this experimentation exhibit strong performance under optimum illumination and environmental conditions. However, our results point toward a significant performance advantage for cluster-based anomaly detectors in the presence of adverse environmental conditions.
The use of hyperspectral imaging (HSI) technology to support a variety of civilian, commercial, and military remote
sensing applications, is growing. The rich spectral information present in HSI allows for more accurate ground cover
identification and classification than with panchromatic or multispectral imagery. One class of problems where
hyperspectral images can be exploited, even when no a priori information about a particular ground cover class is
available, is anomaly detection. Here spectral outliers (anomalies) are detected based on how well each hyperpixel
(spectral irradiance vector for a given pixel position) fits within some background statistical model. Spectral anomalies
may correspond to areas of interest in a given scene. In this work, we compare several anomaly detectors found in the
literature in novel experiments. In particular, we study the performance of the anomaly detectors in detecting several
man-made painted panels in a natural background using visible/near-infrared hyperspectral imagery. The data have been
collected over the course of a nine month period, allowing us to test the robustness of the anomaly detectors with
seasonal change. The detectors considered include the simple Gaussian anomaly detector, a Gaussian mixture model
(GMM) anomaly detector, and the cluster-based anomaly detector (CBAD). We examine the effect of the number of
components for the GMM and the number of clusters for the CBAD. Our preliminary results suggest that the use of a
CBAD yields the best results for our data.